WO2019221581A1 - Exhalation-based lung cancer diagnosis method and system - Google Patents

Exhalation-based lung cancer diagnosis method and system Download PDF

Info

Publication number
WO2019221581A1
WO2019221581A1 PCT/KR2019/006018 KR2019006018W WO2019221581A1 WO 2019221581 A1 WO2019221581 A1 WO 2019221581A1 KR 2019006018 W KR2019006018 W KR 2019006018W WO 2019221581 A1 WO2019221581 A1 WO 2019221581A1
Authority
WO
WIPO (PCT)
Prior art keywords
sers
signal
lung cancer
exhalation
cell
Prior art date
Application number
PCT/KR2019/006018
Other languages
French (fr)
Korean (ko)
Inventor
최연호
성준경
김현구
심온
신현구
Original Assignee
고려대학교 산학협력단
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 고려대학교 산학협력단 filed Critical 고려대학교 산학협력단
Priority to CN201980032939.7A priority Critical patent/CN112136036B/en
Priority to US17/054,270 priority patent/US20210247382A1/en
Priority to EP19803582.6A priority patent/EP3795983A4/en
Priority claimed from KR1020190058785A external-priority patent/KR102225543B1/en
Publication of WO2019221581A1 publication Critical patent/WO2019221581A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/65—Raman scattering
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/483—Physical analysis of biological material
    • G01N33/497—Physical analysis of biological material of gaseous biological material, e.g. breath
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present invention relates to a technique for diagnosing lung cancer based on exhalation, and relates to an exhalation-based lung cancer diagnosis method for diagnosing lung cancer by analyzing a difference between VOC components of exhalation of a lung cancer patient and a normal person.
  • Lung cancer is often diagnosed using computed tomography (CT) and biopsies where invasive processes are essential.
  • CT computed tomography
  • the present invention analyzes the difference between the VOC components of the exhalation of lung cancer patients and normal people by using deep learning, which is a classification method using SERS and artificial intelligence, to diagnose lung cancer quickly and accurately It relates to an exhalation-based lung cancer diagnostic method that can be performed.
  • a surface enhanced Raman Spectroscopy (SERS) substrate Eluting a VOC (Volatile Organic Compound) contained in each of a plurality of cells, supplying each of the cell VOC eluates to the SERS substrate and measuring each cell SERS signal; Deep learning each cell SERS signal to learn a signal pattern of each cell; Capturing a patient breath and liquefying with silicone oil, supplying the liquefied patient breath to the SERS substrate, measuring a breath SERS signal, and analyzing the deep learning result to determine a signal pattern of the breath; And comparing and analyzing signal patterns of the cellular SERS signals and signal patterns of the exhaled SERS signals to identify the cellular SERS signals having the highest similarity to the exhaled SERS signals, and confirming and notifying the presence or absence of lung cancer cells based thereon. It can provide a breath-based lung cancer diagnostic method comprising the step of.
  • VOC Volatile Organic Compound
  • the plurality of cells are composed of lung cancer cells and normal cells
  • the lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, At least one of N417
  • the normal cell is characterized in that the Alveolar type II cell.
  • the measuring of each of the cell SERS signals may be performed by injecting silicone oil into a culture medium in which the cells to be learned are cultured and stirred for a predetermined time, so that the VOCs contained in the cells to be learned are eluted in the silicon oil.
  • Measuring the exhaled SERS signal comprises collecting a patient exhaled in a tedler bag, injecting silicone oil into the tethered bag and waiting for a predetermined time, thereby reducing the VOC contained in the exhaled patient. It is characterized in that to be eluted to the oil.
  • the step of learning the signal pattern of each of the cells learning the signal pattern of each of the cell SERS signal on the basis of dictionary learning the step of identifying the signal pattern of the exhalation is the signal pattern of the exhalation SERS signal on the basis of dictionary learning It is characterized by grasping.
  • the checking and notifying of the presence of lung cancer cells detects the cell SERS signal having the highest similarity through correlation analysis between the signal pattern of each of the cellular SERS signals and the signal pattern of the exhaled SERS signal, and detects the detected cells. According to the type of the SERS signal characterized in that the presence of lung cancer cells.
  • Learning the signal pattern of each cell is to learn the correlation between the cell SERS signal and the cell type through CNN (Convolutional Neural Networks) consisting of one input layer, a plurality of hidden layers, one output layer, Acquiring a signal pattern of each of the cellular SERS signals through a CNN, and identifying a signal pattern of the exhalation, while determining a signal pattern of the exhaled SERS signal through the learned CNN, wherein each of the cellular SERS signals and the exhalation
  • the signal pattern of the SERS signal is determined based on the i (i is a natural number of 2 or more) output information of the hidden layer.
  • the detecting and notifying of the presence of lung cancer cells may include detecting a cell SERS signal having the highest similarity based on a distance value between a signal pattern of each of the cell SERS signals and a signal pattern of the exhaled SERS signal, and detecting the detected cells. According to the type of the SERS signal characterized in that the presence of lung cancer cells.
  • SERS Surface Enhanced Raman Spectroscopy
  • a Raman spectrometer for measuring cellular SERS signals when a VOC (Volatile Organic Compound) eluate of each of a plurality of cells is supplied to the SERS substrate, and measuring an expiratory SERS signal when an exhaled VOC eluate is supplied to the SERS substrate;
  • VOC Volatile Organic Compound
  • a deep learning unit configured to acquire and output a signal pattern of the exhaled SERS signal;
  • comparing and analyzing a signal pattern of the exhaled SERS signal and a signal pattern of the cellular SERS signals to detect cellular SERS signals having the highest similarity, and to determine whether lung cancer cells exist according to the type of the detected cellular SERS signals.
  • the cell VOC eluate is obtained by injecting silicone oil into the culture medium in which the cells are cultured and stirring for a predetermined time, and then separating the silicone oil from which the VOC contained in the cell is eluted.
  • the plurality of cells may be lung cancer cells and normal cells
  • the lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, At least one of N417
  • the normal cell is characterized in that the Alveolar type II cell.
  • the exhaled VOC eluate is obtained by capturing a patient exhalation in a tedler bag, injecting silicone oil into the tedler bag and waiting for a preset time, after which the VOC contained in the exhalation of the patient separates the silicone oil. It is characterized by possible.
  • the deep learning unit may acquire and store a signal pattern of an input signal by deep learning on the basis of dictionary learning.
  • the deep learning unit acquires and stores a signal pattern of each input signal by deep learning based on CNN (Convolutional Neural Networks) including one input layer, a plurality of hidden layers, and one output layer, and the signal pattern is a i (i is a natural number of 2 or more) is determined based on the output information.
  • CNN Convolutional Neural Networks
  • the exhalation-based lung cancer diagnosis method of the present invention can be performed non-invasively by diagnosing lung cancer by analyzing the difference between the VOC components of the exhalation of lung cancer patients and normal people, and also prevents the possibility of radiation exposure in advance.
  • FIG 1 and 2 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to an embodiment of the present invention.
  • FIG. 3 is a view for explaining in more detail the SERS substrate manufacturing step according to an embodiment of the present invention.
  • Figure 4 is a view for explaining the cell SERS signal measuring step according to an embodiment of the present invention in more detail.
  • FIG. 5 illustrates Raman spectra of cellular SERS signals according to an embodiment of the present invention.
  • FIG. 6 is a view for explaining a deep learning method according to an embodiment of the present invention.
  • FIG. 7 is a view for explaining the patient exhalation liquefaction step according to an embodiment of the present invention in more detail.
  • FIG. 8 is a view for explaining in detail the step of confirming the presence of lung cancer cells according to an embodiment of the present invention.
  • FIGS 9 and 10 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to another embodiment of the present invention.
  • FIG. 11 is a view for explaining a breath-based lung cancer diagnostic system according to an embodiment of the present invention.
  • FIG 1 and 2 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to an embodiment of the present invention.
  • the exhalation-based lung cancer diagnostic method of the present invention to prepare a surface enhanced Raman Spectroscopy (SERS) substrate (S1), eluting a VOC (Volatile Organic Compound) contained in each of the cells to be studied Injecting an eluate of each of the cells to be studied to each of the SERS substrate and measuring the SERS signal (S2), by deep learning the SERS signal of each of the cells to be learned specificity of each of the cells to be learned Learning the red SERS peak (S3), collecting the patient's exhalation, and liquefying through the silicone oil (S4), and injecting the liquefied patient's exhalation onto the SERS substrate and measuring the SERS signal (S5). And comparing and analyzing the SERS signal of the patient exhalation with the deep learning result to confirm and notify the presence of lung cancer cells (S6).
  • SERS surface enhanced Raman Spectroscopy
  • the present invention is to diagnose the lung cancer of the patient by a new respiratory analysis method using deep learning, which is a classification method using SERS technology and artificial intelligence.
  • SERS is obtained by amplifying a unique SERS signal on the surface of a nanostructure, which occurs differently depending on the vibration state of molecules when light enters a material.
  • SERS signals specific to each substance have been recently used to qualitatively measure biological substances.
  • the conventional SERS is mainly used only in the liquid or solid phase, and rarely used in the gas phase, because the SERS is affected by the state of molecular vibration.
  • the VOC contained in the patient's exhalation is eluted in a solvent such as silicone oil, and the SERS can be performed based on the eluate.
  • the present invention when analyzing the patient's breathing immediately, in consideration of the difficulty in distinguishing the differences caused by the surrounding factors such as smoking, ambient air quality, the present invention is purely specific VOC SERS peak that occurs in lung cancer cells Acquire and analyze SERS signals for multiple lung cancer cells that can represent a variety of actual lung carcinomas.
  • NSCLC Non-Small Cell Lung Cancer
  • this may also be variously applicable.
  • the SERS substrate fabrication step of step S1 may be performed as in FIG. 3.
  • a base substrate is prepared (S11), and the base substrate surface is coated with a silane coupling agent and then washed (S12).
  • the base substrate may be implemented with a cover glass or the like.
  • the silane coupling agent is a material for enhancing the bonding strength between the cover glass and the nanoparticles, and may be implemented with 3-APTES (aminopropyltriethoxysilane) at a concentration of 0.1%.
  • the base substrate in the dodecanethiol solution for example, a predetermined time (for example, 72 hours) and soaked to coat the surface of the gold nanoparticles (S14).
  • the size of the gold nanoparticles is preferably about 80 nm, and the dodecanethiol solution is preferably ethanol as a solvent and has a concentration of about 0.1%.
  • the cell SERS signal measuring step of step S2 may be performed as in FIG. 4.
  • each of the cells to be studied is cultured for a preset time (eg, 72 hours) (S21).
  • the target cells are composed of lung cancer cells and normal cells
  • lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 and the like
  • normal cells may be Alveolar type II cells and the like.
  • silicon oil is injected into the cell culture solution, and the layered cell culture solution and the silicone oil are stirred for a predetermined time (for example, 30 minutes) or more to elute the VOC contained in the cell culture (S22).
  • the SERS substrate After separating the cell eluate and supplying it to the SERS substrate (S23), the SERS substrate is irradiated with a 785 nm laser to measure the SERS signal (S24).
  • the SERS signals obtained through step S24 may have a Raman spectrum as shown in FIG. 5. Referring to this, it can be seen that all of the SERS signals corresponding to each cell have different signal patterns.
  • the present invention seeks to overcome the heterogeneity between patients by measuring VOCs of lung cancer cell lines and obtaining dictionary elements based on them, that is, by using VOCs obtained from pure cancer cells.
  • a signal pattern ie, a specific SERS peak
  • dictionary learning is a kind of analysis method using artificial intelligence and is performed as follows.
  • a specific database is represented as a dictionary, which is a linear combination of basic elements called atoms.
  • atom represents m signal characteristics derived from each of the signals included in the database as a column vector (m ⁇ 1), and hereinafter, the description will be referred to as a dictionary element.
  • Dictionary learning is an example of deep learning, where a dictionary element can be described generalized to the signal pattern of the SERS signal.
  • the present invention deep learning the SERS signal (X) of each of the lung cancer cells and normal cells to acquire and store a signal pattern corresponding to each of the cell SERS signals.
  • Patient exhalation liquefaction step of step S4 may be performed as in FIG.
  • Tedler A syringe is inserted into the bag to allow the VOC contained in the patient's exhalation to remove the eluted silicone oil (ie, the exhalation eluent) from the Tedlar bag (S43).
  • the SERS substrate of the exhalation is measured by irradiating the SERS substrate with a 785 nm laser.
  • step S6 deep breathing of the exhaled SERS signal, a deep dictionary run on the exhaled SERS signal is performed to acquire a signal pattern of exhalation (ie, a patient-derived SERS signal pattern).
  • step S7 through correlation analysis between the signal pattern of the exhaled SERS signal and the signal pattern learned through step S3, as shown in FIG. Detect the cells.
  • the red box in FIG. 8 shows a pair of cells having the highest correlation between the actual patient-derived SERS signal pattern and the SERS signal pattern learned through deep learning.
  • lung cancer may be diagnosed and notified only if the detected cells are lung cancer cells.
  • dictionary learning has been described as an example of deep learning, but the present invention may also support analysis operations based on a convolutional neural network (CNN).
  • CNN convolutional neural network
  • FIGS. 9 and 10 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to another embodiment of the present invention.
  • This is a CNN-based method instead of dictionary learning, and the steps related to SERS substrate, SERS signal measurement, etc. are performed in the same manner as the dictionary learning based method, and a detailed description thereof will be omitted.
  • CNN is composed of one input layer, multiple hidden layers, and one output layer, and neurons included in each layer are connected by weight, so that multiple learning with cell SERS signal as input condition and cell type as output condition Generate and train the data to adjust the neuron weights included in each layer.
  • the CNN is used to determine the signal pattern of the cellular SERS signal or the exhaled SERS signal, and in particular, i (i is 2 or more natural numbers, for example, 10) of the hidden layer instead of the output layer for notifying the cell type.
  • the output information is acquired and utilized as a signal pattern corresponding to the input signal.
  • SERS signals ie, cellular SERS signals
  • lung cancer cells and normal cells are acquired.
  • CNN model is repeatedly trained through them, and CNN model subsequently classifies input signals into lung cancer cells and normal cells. classify.
  • the CNN model is trained, i (e.g., 10) hidden layer output information for each of the cellular SERS signals are obtained and stored as a signal pattern of each of the cellular SERS signals (S3 ').
  • step S3 ' when the aerobic SERS signal is obtained (S4, S5), the aerobic SERS signal is inputted into the trained CNN model, and then i (e.g., 10) hidden layers as in the cellular SERS signal.
  • the output information is acquired and stored as a signal pattern of the exhalation SERS signal (S6 ').
  • the distance between the signal pattern of each of the cellular SERS signal and the signal pattern of the aerobic SERS signal is calculated to derive and notify the cells having the highest similarity to the exhalation signal based on the cellular SERS signal having the shortest distance value.
  • the distance calculation method may be any one of Mahalanobis distance, Euclidean distance, cosine distance, etc., and the similarity is increased in inverse proportion to the distance.
  • FIG. 11 is a view for explaining a breath-based lung cancer diagnostic system according to an embodiment of the present invention.
  • the system of the present invention includes a SERS substrate 10, a Raman spectrometer 20 for measuring SERS signals for cellular VOC (Volatile Organic Compound) eluates and aerobic VOC eluates supplied to each of the SERS substrates,
  • the deep learning unit 30 acquires and stores the pattern of the cellular SERS signals, and when the exhaled SERS signal is measured, the deep learning unit 30 that acquires and outputs the pattern of the exhaled SERS signals,
  • the cell SERS signal having the highest similarity is detected, and the presence or absence of lung cancer cells is determined according to the type of the detected cell SERS signal.
  • Lung cancer diagnosis unit 40 and the like are performed by performing correlation analysis between the pattern of the exhaled SERS signal and the pattern learned by the deep learning.
  • the cell VOC eluate can be obtained by injecting silicone oil into the culture medium in which the cells are cultured and stirring for a predetermined time, and then separating the silicone oil eluted with VOC contained in the cells.
  • Exhaled VOC eluate is also obtained by capturing the patient exhalation in a tedler bag, injecting silicone oil into the tedler bag and waiting for a preset time, after which the VOC contained in the patient exhalation separates the silicone oil. It is possible.
  • the cells at this time may also be lung cancer cells and normal cells as described above, lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, It may be at least one of EBC1, H1650, N417, the normal cell may be an Alveolar type II cell.
  • the system of the present invention also uses deep learning (ie, dictionary learning, CNN), which is a classification method using SERS technology and artificial intelligence described above, and thus can perform lung cancer diagnosis operation without worrying about invasive radiation exposure.
  • lung cancer based on lung cancer cells and normal cells, but if necessary, bronchial cancer, colorectal cancer, prostate cancer, breast cancer, pancreatic cancer, gastric cancer, ovarian cancer, bladder cancer, brain cancer, thyroid cancer, Various cancers other than lung cancer may also be diagnosed based on cancer cells corresponding to esophageal cancer, uterine cancer, liver cancer, kidney cancer, biliary tract cancer, and testicular cancer.

Abstract

The present invention relates to an exhalation-based lung cancer diagnosis method and system which may comprise the steps of: producing a surface enhanced Raman spectroscopy (SERS) substrate; eluting volatile organic compounds (VOCs) included in each of a plurality of cells, supplying cell VOC eluates to the SERS substrate, and then measuring respective cell SERS signals thereof; learning signal patterns of each of the cells by applying deep-learning to each of the cell SERS signals; collecting patient exhalation and liquefying same by silicone oil, and supplying the liquefied patient exhalation to the SERS substrate and measuring an exhalation SERS signal, and then identifying the signal pattern of the exhalation by analyzing the exhalation SERS signal through the deep-learning result; and comparing and analyzing the signal patterns of each of the cell SERS signals with the signal pattern of the exhalation SERS signal, thereby identifying the cell SERS signal having the highest similarity to the exhalation SERS signal, and on the basis thereof, confirming whether a lung cancer cell is present or not, and notifying thereof.

Description

호기 기반 폐암 진단 방법 및 시스템Exhalation-based lung cancer diagnosis method and system
본 발명은 호기 기반으로 폐암을 진단하기 위한 기술에 관한 것으로, 폐암 환자와 정상인의 호기의 VOC 성분의 차이를 분석하는 방식으로 폐암을 진단할 수 있도록 하는 호기 기반 폐암 진단 방법에 관한 것이다. The present invention relates to a technique for diagnosing lung cancer based on exhalation, and relates to an exhalation-based lung cancer diagnosis method for diagnosing lung cancer by analyzing a difference between VOC components of exhalation of a lung cancer patient and a normal person.
폐암은 CT(Computed Tomography)와 침습적인 과정이 필수적인 생검을 이용하여 주로 진단된다. Lung cancer is often diagnosed using computed tomography (CT) and biopsies where invasive processes are essential.
그러나 CT의 경우 방사선 피폭의 위험이 있고, 또한 대형 병원에서만 검사가 가능하기 때문에 진단에 어려움이 있었다. 폐암을 간편하게 검사할 수 없다는 단점 때문에 폐암의 조기진단이 쉽게 이루어지지 않는다는 점 또한 폐암의 높은 사망률의 원인으로 꼽힌다. However, in the case of CT, there is a risk of radiation exposure and it is difficult to diagnose because the test can be performed only in a large hospital. Due to the disadvantage that lung cancer cannot be easily examined, the early diagnosis of lung cancer is not easy.
이러한 단점을 보완하기 위해 간편하고 비침습적인 호기 가스를 이용한 폐암 진단법이 연구되어 왔다. 호흡 기반 폐암 진단법에서는 GC-MS 기반의 분석법을 통해 호흡 내 VOC의 바이오마커를 진단하는 데, 이는 다음의 문제점을 가진다. In order to make up for this drawback, lung cancer diagnosis using simple and non-invasive exhalation gas has been studied. In the diagnosis of respiratory lung cancer, the biomarker of VOC in respiration is diagnosed through GC-MS-based analysis, which has the following problems.
첫째, 환자의 흡연 유무나 성별 등에 의해 결과가 달라졌다는 점이고, 둘째, GC-MS 분석결과에서 나타나는 다양성(heterogenicity)이다. 폐암 환자와 정상에서 차이를 보이는 물질은 에틸 알코올(ethyl alcohol)과 같은 특정한 개개의 물질이 아닌 알데히드(aldehyde), 알케인(alkane)과 같은 물질 군으로 나타났다는 점이었다. 따라서 GC-MS를 통해서는 호흡 VOC의 폐암 특이적 바이오마커를 선정할 수 없는 문제가 있다. First, the results were different according to the patient's smoking status and gender, and second, the heterogeneity in the GC-MS analysis results. The difference between normal and lung cancer patients was that they appeared in groups of aldehydes and alkanes, rather than specific individual substances such as ethyl alcohol. Therefore, there is a problem in that lung cancer specific biomarkers of respiratory VOC cannot be selected through GC-MS.
이에 상기와 같은 문제점을 해결하기 위한 것으로서, 본 발명은 폐암 환자와 정상인의 호기의 VOC 성분의 차이를 SERS과 인공지능을 이용한 분류 방법인 딥 러닝을 이용하여 분석함으로써, 사용자 호기만으로 신속 정확한 폐암 진단을 수행할 수 있도록 하는 호기 기반 폐암 진단 방법에 관한 것이다. In order to solve the above problems, the present invention analyzes the difference between the VOC components of the exhalation of lung cancer patients and normal people by using deep learning, which is a classification method using SERS and artificial intelligence, to diagnose lung cancer quickly and accurately It relates to an exhalation-based lung cancer diagnostic method that can be performed.
본 발명의 목적은 이상에서 언급한 목적으로 제한되지 않으며, 언급되지 않은 또 다른 목적들은 아래의 기재로부터 본 발명이 속하는 통상의 지식을 가진 자에게 명확하게 이해될 수 있을 것이다.The object of the present invention is not limited to the above-mentioned object, and other objects which are not mentioned will be clearly understood by those skilled in the art from the following description.
상기 과제를 해결하기 위한 수단으로서, 본 발명의 일 실시 형태에 따르면 SERS(Surface Enhanced Raman Spectroscopy) 기판을 제작하는 단계; 다수의 세포 각각에 포함된 VOC(Volatile Organic Compound)를 용출하고, 세포 VOC 용출액 각각을 상기 SERS 기판에 공급한 후 세포 SERS 신호 각각을 측정하는 단계; 상기 세포 SERS 신호 각각을 딥 러닝하여 상기 세포 각각의 신호 패턴을 학습하는 단계; 환자 호기를 포집한 후 실리콘 오일을 통해 액화시키고, 상기 액화된 환자 호기를 상기 SERS 기판에 공급하여 호기 SERS 신호를 측정한 후, 상기 딥 러닝 결과를 통해 분석하여 호기의 신호 패턴을 파악하는 단계; 및 상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴을 비교 분석하여, 상기 호기 SERS 신호와 가장 유사도가 높은 상기 세포 SERS 신호를 파악하고, 이를 기반으로 폐암 세포 존재 여부를 확인 및 통보하는 단계를 포함하는 호기 기반 폐암 진단 방법을 제공할 수 있다. As a means for solving the above problems, according to an embodiment of the present invention manufacturing a surface enhanced Raman Spectroscopy (SERS) substrate; Eluting a VOC (Volatile Organic Compound) contained in each of a plurality of cells, supplying each of the cell VOC eluates to the SERS substrate and measuring each cell SERS signal; Deep learning each cell SERS signal to learn a signal pattern of each cell; Capturing a patient breath and liquefying with silicone oil, supplying the liquefied patient breath to the SERS substrate, measuring a breath SERS signal, and analyzing the deep learning result to determine a signal pattern of the breath; And comparing and analyzing signal patterns of the cellular SERS signals and signal patterns of the exhaled SERS signals to identify the cellular SERS signals having the highest similarity to the exhaled SERS signals, and confirming and notifying the presence or absence of lung cancer cells based thereon. It can provide a breath-based lung cancer diagnostic method comprising the step of.
상기 다수의 세포는 폐암 세포와 정상 세포로 구성되며, 상기 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 중 적어도 하나일 수 있으며, 상기 정상 세포는 Alveolar type II cell인 것을 특징으로 한다. The plurality of cells are composed of lung cancer cells and normal cells, the lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, At least one of N417, the normal cell is characterized in that the Alveolar type II cell.
상기 세포 SERS 신호 각각을 측정하는 단계는 학습 대상 세포를 배양한 배양액에 실리콘 오일을 주입하고 기 설정 시간 동안 교반하여, 상기 학습 대상 세포에 포함된 VOC가 실리콘 오일에 용출되도록 하는 것을 특징으로 한다. The measuring of each of the cell SERS signals may be performed by injecting silicone oil into a culture medium in which the cells to be learned are cultured and stirred for a predetermined time, so that the VOCs contained in the cells to be learned are eluted in the silicon oil.
상기 호기 SERS 신호를 측정하는 단계는 환자 호기를 테들러 백(tedler bag)에 포집한 후, 상기 테들러 백에 실리콘 오일을 주입하고 기 설정 시간 동안 기다림으로써, 상기 환자 호기에 포함된 VOC가 실리콘 오일에 용출되도록 하는 것을 특징으로 한다. Measuring the exhaled SERS signal comprises collecting a patient exhaled in a tedler bag, injecting silicone oil into the tethered bag and waiting for a predetermined time, thereby reducing the VOC contained in the exhaled patient. It is characterized in that to be eluted to the oil.
더하여, 상기 세포 각각의 신호 패턴을 학습하는 단계는 딕셔너리 러닝 기반으로 상기 세포 SERS 신호 각각의 신호 패턴을 학습하고, 상기 호기의 신호 패턴을 파악하는 단계는 딕셔너리 러닝 기반으로 상기 호기 SERS 신호의 신호 패턴을 파악하는 것을 특징으로 한다. In addition, the step of learning the signal pattern of each of the cells learning the signal pattern of each of the cell SERS signal on the basis of dictionary learning, the step of identifying the signal pattern of the exhalation is the signal pattern of the exhalation SERS signal on the basis of dictionary learning It is characterized by grasping.
상기 폐암 세포 존재 여부를 확인 및 통보하는 단계는 상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴에 대한 상관 분석을 통해 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 것을 특징으로 한다. The checking and notifying of the presence of lung cancer cells detects the cell SERS signal having the highest similarity through correlation analysis between the signal pattern of each of the cellular SERS signals and the signal pattern of the exhaled SERS signal, and detects the detected cells. According to the type of the SERS signal characterized in that the presence of lung cancer cells.
상기 세포 각각의 신호 패턴을 학습하는 단계는 하나의 입력층, 다수의 은닉층, 하나의 출력층으로 구성되는 CNN(Convolutional Neural Networks)를 통해 세포 SERS 신호와 세포 종류간의 상관 관계를 학습시킨 후, 학습 완료된 CNN를 통해 상기 세포 SERS 신호 각각의 신호 패턴을 획득하고, 상기 호기의 신호 패턴을 파악하는 단계는 상기 학습 완료된 CNN를 통해 상기 호기 SERS 신호의 신호 패턴을 파악하되, 상기 세포 SERS 신호 각각과 상기 호기 SERS 신호의 신호 패턴은 상기 은닉층의 i개(i는 2 이상의 자연수) 출력 정보에 기반하여 결정되는 것을 특징으로 한다. Learning the signal pattern of each cell is to learn the correlation between the cell SERS signal and the cell type through CNN (Convolutional Neural Networks) consisting of one input layer, a plurality of hidden layers, one output layer, Acquiring a signal pattern of each of the cellular SERS signals through a CNN, and identifying a signal pattern of the exhalation, while determining a signal pattern of the exhaled SERS signal through the learned CNN, wherein each of the cellular SERS signals and the exhalation The signal pattern of the SERS signal is determined based on the i (i is a natural number of 2 or more) output information of the hidden layer.
상기 폐암 세포 존재 여부를 확인 및 통보하는 단계는 상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴간 거리값에 기반하여 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 것을 특징으로 한다. The detecting and notifying of the presence of lung cancer cells may include detecting a cell SERS signal having the highest similarity based on a distance value between a signal pattern of each of the cell SERS signals and a signal pattern of the exhaled SERS signal, and detecting the detected cells. According to the type of the SERS signal characterized in that the presence of lung cancer cells.
상기 과제를 해결하기 위한 수단으로서, 본 발명의 다른 실시 형태에 따르면 SERS(Surface Enhanced Raman Spectroscopy) 기판; 다수의 세포 각각의 VOC(Volatile Organic Compound) 용출액이 상기 SERS 기판에 공급되는 경우에는 세포 SERS 신호들을 측정하고, 호기 VOC 용출액이 상기 SERS 기판에 공급되는 경우에는 호기 SERS 신호를 측정하는 라만 분광기; 상기 라만 분광기에 의해 세포 SERS 신호들이 측정되면, 상기 세포 SERS 신호들에 기반한 딥 러닝을 수행하여 상기 세포 SERS 신호들의 신호 패턴을 획득 및 저장하고, 호기 SERS 신호가 측정되면, 상기 립 러닝 결과를 통해 상기 호기 SERS 신호의 신호 패턴을 획득 및 출력하는 딥 러닝부; 및 상기 호기 SERS 신호의 신호 패턴과 상기 세포 SERS 신호들의 신호 패턴을 비교 분석하여, 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 폐암 진단부를 포함하는 호기 기반 폐암 진단 시스템을 제공한다. As a means for solving the above problems, according to another embodiment of the present invention; Surface Enhanced Raman Spectroscopy (SERS) substrate; A Raman spectrometer for measuring cellular SERS signals when a VOC (Volatile Organic Compound) eluate of each of a plurality of cells is supplied to the SERS substrate, and measuring an expiratory SERS signal when an exhaled VOC eluate is supplied to the SERS substrate; When the cellular SERS signals are measured by the Raman spectrometer, deep learning based on the cellular SERS signals is performed to acquire and store signal patterns of the cellular SERS signals, and when the exhaled SERS signal is measured, the lip learning results A deep learning unit configured to acquire and output a signal pattern of the exhaled SERS signal; And comparing and analyzing a signal pattern of the exhaled SERS signal and a signal pattern of the cellular SERS signals to detect cellular SERS signals having the highest similarity, and to determine whether lung cancer cells exist according to the type of the detected cellular SERS signals. It provides an exhalation-based lung cancer diagnostic system comprising a lung cancer diagnostic unit.
상기 세포 VOC 용출액은 세포를 배양한 배양액에 실리콘 오일을 주입하고 기 설정 시간 동안 교반한 후 상기 세포에 포함된 VOC가 용출된 실리콘 오일을 분리시킴으로써 획득 가능한 것을 특징으로 한다. The cell VOC eluate is obtained by injecting silicone oil into the culture medium in which the cells are cultured and stirring for a predetermined time, and then separating the silicone oil from which the VOC contained in the cell is eluted.
상기 다수의 세포는 폐암 세포와 정상 세포일 수 있으며, 상기 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 중 적어도 하나일 수 있으며, 상기 정상 세포는 Alveolar type II cell인 것을 특징으로 한다. The plurality of cells may be lung cancer cells and normal cells, the lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, At least one of N417, the normal cell is characterized in that the Alveolar type II cell.
상기 호기 VOC 용출액은 환자 호기를 테들러 백(tedler bag)에 포집하고, 상기 테들러 백에 실리콘 오일을 주입하고 기 설정 시간 동안 기다린 후, 상기 환자 호기에 포함된 VOC가 실리콘 오일을 분리시킴으로써 획득 가능한 것을 특징으로 한다. The exhaled VOC eluate is obtained by capturing a patient exhalation in a tedler bag, injecting silicone oil into the tedler bag and waiting for a preset time, after which the VOC contained in the exhalation of the patient separates the silicone oil. It is characterized by possible.
상기 딥 러닝부는 딕셔너리 러닝 기반으로 딥 러닝하여 입력 신호의 신호 패턴을 획득 및 저장하는 것을 특징으로 한다. The deep learning unit may acquire and store a signal pattern of an input signal by deep learning on the basis of dictionary learning.
또한 상기 딥 러닝부는 하나의 입력층, 다수의 은닉층, 하나의 출력층으로 구성되는 CNN(Convolutional Neural Networks) 기반으로 딥 러닝하여 입력 신호 각각의 신호 패턴을 획득 및 저장하되, 상기 신호 패턴은 상기 은닉층의 i개(i는 2 이상의 자연수) 출력 정보에 기반하여 결정되는 것을 특징으로 한다. In addition, the deep learning unit acquires and stores a signal pattern of each input signal by deep learning based on CNN (Convolutional Neural Networks) including one input layer, a plurality of hidden layers, and one output layer, and the signal pattern is a i (i is a natural number of 2 or more) is determined based on the output information.
본 발명의 호기 기반 폐암 진단 방법은 폐암 환자와 정상인의 호기의 VOC 성분의 차이를 분석하는 방식으로 폐암 여부를 진단하도록 함으로써 비침습적으로 수행될 수 있으며 방사선 노출 가능성도 사전 차단하도록 한다. The exhalation-based lung cancer diagnosis method of the present invention can be performed non-invasively by diagnosing lung cancer by analyzing the difference between the VOC components of the exhalation of lung cancer patients and normal people, and also prevents the possibility of radiation exposure in advance.
또한, 순수 암 세포에 포함된 VOC를 SERS 측정하고 딥러닝한 후, 이를 활용하여 폐암 여부를 진단하도록 함으로써, 환자간의 다양성(heterogenicity)을 용이하게 극복할 수 있도록 한다. In addition, by measuring the SERS and deep learning of the VOC contained in the pure cancer cells, by using this to diagnose lung cancer, it is possible to easily overcome the heterogeneity (heterogenicity) between patients.
더하여, 호기를 액화한 후 SERS 신호를 측정하는 데 30분 정도의 시간만이 소요되므로, 평균 4시간 이상 소요되는 GC-MS기법에 비해 매우 빠른 진단 시간을 제공할 수 있다. In addition, since it takes only about 30 minutes to measure the SERS signal after liquefying the exhalation, it can provide a very fast diagnostic time compared to the GC-MS technique that takes an average of 4 hours or more.
도 1 및 도 2은 본 발명의 일 실시예에 따른 호기 기반 폐암 진단 방법을 개략적으로 설명하기 위한 도면이다. 1 and 2 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to an embodiment of the present invention.
도 3은 본 발명의 일 실시예에 따른 SERS 기판 제작 단계를 보다 상세히 설명하기 위한 도면이다. 3 is a view for explaining in more detail the SERS substrate manufacturing step according to an embodiment of the present invention.
도 4는 본 발명의 일 실시예에 따른 세포 SERS 신호 측정 단계를 보다 상세히 설명하기 위한 도면이다. Figure 4 is a view for explaining the cell SERS signal measuring step according to an embodiment of the present invention in more detail.
도 5는 본 발명의 일 실시예에 따른 세포 SERS 신호들의 라만 스펙트럼을 도시한 도면이다. 5 illustrates Raman spectra of cellular SERS signals according to an embodiment of the present invention.
도 6은 본 발명의 일 실시예에 따른 딥 러닝 방법을 설명하기 위한 도면이다. 6 is a view for explaining a deep learning method according to an embodiment of the present invention.
도 7는 본 발명의 일 실시예에 따른 환자 호기 액화 단계를 보다 상세히 설명하기 위한 도면이다. 7 is a view for explaining the patient exhalation liquefaction step according to an embodiment of the present invention in more detail.
도 8는 본 발명의 일 실시예에 따른 폐암 세포 존재 여부 확인 단계를 보다 상세히 설명하기 위한 도면이다. 8 is a view for explaining in detail the step of confirming the presence of lung cancer cells according to an embodiment of the present invention.
도 9 및 도 10은 본 발명의 다른 실시예에 따른 호기 기반 폐암 진단 방법을 개략적으로 설명하기 위한 도면이다. 9 and 10 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to another embodiment of the present invention.
도 11은 본 발명의 일 실시예에 따른 호기 기반 폐암 진단 시스템을 설명하기 위한 도면이다. 11 is a view for explaining a breath-based lung cancer diagnostic system according to an embodiment of the present invention.
본 발명의 목적 및 효과, 그리고 그것들을 달성하기 위한 기술적 구성들은 첨부되는 도면과 함께 상세하게 후술되어 있는 실시예들을 참조하면 명확해질 것이다. 본 발명을 설명함에 있어서 공지 기능 또는 구성에 대한 구체적인 설명이 본 발명의 요지를 불필요하게 흐릴 수 있다고 판단되는 경우에는 그 상세한 설명을 생략할 것이다.The objects and effects of the present invention and the technical configurations for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. In describing the present invention, when it is determined that a detailed description of a known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.
그리고 후술되는 용어들은 본 발명에서의 기능을 고려하여 정의된 용어들로서 이는 사용자, 운용자의 의도 또는 관례 등에 따라 달라질 수 있다.Terms to be described later are terms defined in consideration of functions in the present invention, and may be changed according to intentions or customs of users or operators.
그러나 본 발명은 이하에서 개시되는 실시예들에 한정되는 것이 아니라 서로 다른 다양한 형태로 구현될 수 있다. 단지 본 실시예들은 본 발명의 개시가 완전하도록 하고, 본 발명이 속하는 기술분야에서 통상의 지식을 가진 자에게 발명의 범주를 완전하게 알려주기 위해 제공되는 것이며, 본 발명은 청구항의 범주에 의해 정의될 뿐이다. 그러므로 그 정의는 본 명세서 전반에 걸친 내용을 토대로 내려져야 할 것이다.However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various forms. The present embodiments are merely provided to complete the disclosure of the present invention and to fully inform the scope of the invention to those skilled in the art, and the present invention is defined by the scope of the claims. It will be. Therefore, the definition should be made based on the contents throughout the specification.
도 1 및 도 2은 본 발명의 일 실시예에 따른 호기 기반 폐암 진단 방법을 개략적으로 설명하기 위한 도면이다. 1 and 2 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to an embodiment of the present invention.
도 1에 도시된 바와 같이, 본 발명의 호기 기반 폐암 진단 방법은 SERS(Surface Enhanced Raman Spectroscopy) 기판을 제작하는 단계(S1), 학습 대상 세포들 각각에 포함된 VOC(Volatile Organic Compound)를 용출하고, 상기 학습 대상 세포들 각각의 용출액을 상기 SERS 기판 각각에 분사한 한 후 SERS 신호를 측정하는 단계(S2), 상기 학습 대상 세포들 각각의 SERS 신호를 딥 러닝하여 상기 학습 대상 세포들 각각의 특이적 SERS 피크를 학습하는 단계(S3), 환자 호기를 포집한 후 실리콘 오일을 통해 액화시키는 단계(S4), 상기 액화된 환자 호기를 상기 SERS 기판에 분사한 후 SERS 신호를 측정하는 단계(S5), 및 상기 환자 호기의 SERS 신호를 상기 딥 러닝 결과와 비교 분석함으로써, 폐암 세포 존재 여부를 확인 및 통보하는 단계(S6)를 포함한다. As shown in Figure 1, the exhalation-based lung cancer diagnostic method of the present invention to prepare a surface enhanced Raman Spectroscopy (SERS) substrate (S1), eluting a VOC (Volatile Organic Compound) contained in each of the cells to be studied Injecting an eluate of each of the cells to be studied to each of the SERS substrate and measuring the SERS signal (S2), by deep learning the SERS signal of each of the cells to be learned specificity of each of the cells to be learned Learning the red SERS peak (S3), collecting the patient's exhalation, and liquefying through the silicone oil (S4), and injecting the liquefied patient's exhalation onto the SERS substrate and measuring the SERS signal (S5). And comparing and analyzing the SERS signal of the patient exhalation with the deep learning result to confirm and notify the presence of lung cancer cells (S6).
즉, 본 발명은 SERS 기술과 인공지능을 이용한 분류 방법인 딥 러닝을 이용하는 새로운 호흡 분석 방식으로 환자의 폐암 여부를 진단하도록 한다. That is, the present invention is to diagnose the lung cancer of the patient by a new respiratory analysis method using deep learning, which is a classification method using SERS technology and artificial intelligence.
참고로, SERS는 빛이 물질에 입사할 때 분자의 진동 상태에 따라 다르게 발생하는 고유한 SERS 신호을 나노 구조체의 표면에서 증폭하여 얻는 것이다. 물질마다 가지는 특이적인 SERS 신호은 생체 물질을 정성적으로 측정하는데 최근 많이 이용되고 있다. 다만, 종래의 SERS는 주로 액체나 고체상에서만 사용되고, 기체상에서는 거의 사용되지 않는 데, 그 이유는 SERS가 분자 진동 상태에 영향을 받기 때문이다. For reference, SERS is obtained by amplifying a unique SERS signal on the surface of a nanostructure, which occurs differently depending on the vibration state of molecules when light enters a material. SERS signals specific to each substance have been recently used to qualitatively measure biological substances. However, the conventional SERS is mainly used only in the liquid or solid phase, and rarely used in the gas phase, because the SERS is affected by the state of molecular vibration.
이에 본 발명에서는 환자 호기에 포함된 VOC를 실리콘 오일(silicone oil)와 같은 용매에 용출하고, 그 용출액을 기반으로 SERS이 수행될 수 있도록 한다. Accordingly, in the present invention, the VOC contained in the patient's exhalation is eluted in a solvent such as silicone oil, and the SERS can be performed based on the eluate.
또한 환자의 호흡을 바로 분석할 경우, 흡연 유무, 주위 공기질과 같은 주위 요소에 의한 차이점을 구분하기가 어려움을 고려하여, 본 발명은 순수하게 폐암 세포에서 발생하는 특이적 VOC SERS 피크(peak)를 획득하고, 실제 다양한 폐암종을 대변할 수 있는 여러 폐암 세포에 대한 SERS 신호를 획득하고 분석하도록 한다. In addition, when analyzing the patient's breathing immediately, in consideration of the difficulty in distinguishing the differences caused by the surrounding factors such as smoking, ambient air quality, the present invention is purely specific VOC SERS peak that occurs in lung cancer cells Acquire and analyze SERS signals for multiple lung cancer cells that can represent a variety of actual lung carcinomas.
폐암 환자의 85%는 비소세포 폐암(NSCLC; NonSmall Cell Lung Cancer)에 속하고, 그 중에서도 40%는 선암(adenocarcinoma), 25~30%는 편평 세포암(Squamous cell carcinoma), 15~10%는 대세포암(Large cell carcinoma)에 속한다. 그리고 NSCLC에 속하지 않는 15%의 폐암 환자는 소세포 폐암(SCLC; Small cell lung cancer)에 속한다. 85% of lung cancer patients belong to Non-Small Cell Lung Cancer (NSCLC), 40% of which are adenocarcinoma, 25 to 30% of squamous cell carcinoma, and 15 to 10% of It belongs to large cell carcinoma. And 15% of lung cancer patients who do not belong to NSCLC belong to small cell lung cancer (SCLC).
따라서 본 발명에서는 각각의 폐암종 세포주인 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 등과, 정상 세포인 Alveolar type II cell 등일 수 있다. 물론, 상기의 세포들 이외에 폐암과 정상인을 구별할 수 있다면, 이 또한 다양하게 적용 가능할 것이다. Therefore, in the present invention, each lung carcinoma cell line A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417, etc. II cell and the like. Of course, if it is possible to distinguish between normal cancer and lung cancer in addition to the above cells, this may also be variously applicable.
다만, 이하에서는 설명의 편이를 위해, 폐암 세포로 A549, H2087, H520, H460, H446를 이용하고, 정상 세포로 Alveolar type II cell을 이용하는 경우에 한해 설명하기로 한다. However, hereinafter, for convenience of description, it will be described only in the case of using A549, H2087, H520, H460, H446 as lung cancer cells and Alveolar type II cells as normal cells.
이하, 도 3 내지 도 8을 참고하여 본 발명의 방법을 보다 상세히 설며하기로 한다. Hereinafter, the method of the present invention will be described in more detail with reference to FIGS. 3 to 8.
단계 S1의 SERS 기판 제작 단계는 도 3에서와 같이 수행될 수 있다. The SERS substrate fabrication step of step S1 may be performed as in FIG. 3.
먼저, 베이스 기판을 마련하고(S11), 베이스 기판 표면을 실란 커플링제로 코팅한 후 세척하도록 한다(S12). 베이스 기판은 커버 글라스 등으로 구현될 수 있다. 실란 커플링제는 커버 글라스와 나노 입자의 결합 강도를 증강시키기 위한 물질로, 0.1% 농도의 3-APTES(aminopropyltriethoxysilane) 등으로 구현될 수 있다. First, a base substrate is prepared (S11), and the base substrate surface is coated with a silane coupling agent and then washed (S12). The base substrate may be implemented with a cover glass or the like. The silane coupling agent is a material for enhancing the bonding strength between the cover glass and the nanoparticles, and may be implemented with 3-APTES (aminopropyltriethoxysilane) at a concentration of 0.1%.
그리고 실란 커플링제의 코팅면에 금 나노 입자를 분사한 후 소정 시간 건조시키고(S13), 금 나노 입자가 완전 건조되면, 베이스 기판을 도데칸티올(dodecanethiol) 용액에 기 설정 시간(예를 들어, 72 시간) 이상 담가두어 금 나노 입자의 표면을 코팅시킨다(S14). 금 나노 입자의 크기는 80nm 정도인 것이 바람직하고, 도데칸티올 용액는 에탄올을 용매제로 하며, 0.1% 정도의 농도인 것이 바람직하다. 이러한 과정을 통해 친수성의 금나노 입자 표면은 VOC와 같은 유기 물질에 대한 친화성을 가질 수 있게 된다. And after spraying the gold nanoparticles to the coating surface of the silane coupling agent and dried for a predetermined time (S13), and when the gold nanoparticles are completely dried, the base substrate in the dodecanethiol solution (for example, a predetermined time (for example, 72 hours) and soaked to coat the surface of the gold nanoparticles (S14). The size of the gold nanoparticles is preferably about 80 nm, and the dodecanethiol solution is preferably ethanol as a solvent and has a concentration of about 0.1%. Through this process, the surface of hydrophilic gold nanoparticles can have affinity for organic materials such as VOC.
단계 S2의 세포 SERS 신호 측정 단계는 도 4에서와 같이 수행될 수 있다. The cell SERS signal measuring step of step S2 may be performed as in FIG. 4.
먼저, 학습 대상 세포 각각을 기 설정 시간(예를 들어, 72시간) 동안 배양한다(S21). 이때, 학습 대상 세포는 폐암 세포와 정상 세포로 구성되며, 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 등일 수 있으며, 정상 세포는 Alveolar type II cell 등일 수 있다. 물론, 상기의 세포들 이외에 폐암과 정상인을 구별할 수 있다면, 이 또한 다양하게 적용 가능할 수 있도록 한다. First, each of the cells to be studied is cultured for a preset time (eg, 72 hours) (S21). At this time, the target cells are composed of lung cancer cells and normal cells, lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 and the like, normal cells may be Alveolar type II cells and the like. Of course, if it is possible to distinguish between normal cancer and lung cancer in addition to the above cells, this also makes it possible to apply variously.
그리고 세포 배양액에 실리콘 오일을 주입하고, 층 분리된 세포 배양액과 실리콘 오일을 기 설정 시간(예를 들어, 30분) 이상 교반시켜 세포 배양속에 포함된 VOC를 용출시킨다(S22). Then, silicon oil is injected into the cell culture solution, and the layered cell culture solution and the silicone oil are stirred for a predetermined time (for example, 30 minutes) or more to elute the VOC contained in the cell culture (S22).
그리고 세포 용출액을 분리시켜 SERS 기판에 공급한 후(S23), SERS 기판에 785 nm 레이저를 조사하여 SERS 신호를 측정하도록 한다(S24). After separating the cell eluate and supplying it to the SERS substrate (S23), the SERS substrate is irradiated with a 785 nm laser to measure the SERS signal (S24).
단계 S24를 통해 획득된 SERS 신호들은 도 5와 같은 라만 스펙트럼을 가질 수 있으며, 이를 참고하면, 세포 각각에 대응되는 SERS 신호 모두가 서로 상이한 신호 패턴을 가짐을 알 수 있다. The SERS signals obtained through step S24 may have a Raman spectrum as shown in FIG. 5. Referring to this, it can be seen that all of the SERS signals corresponding to each cell have different signal patterns.
이와 같이 본 발명은 폐암 세포주의 VOC를 측정하고, 이를 기반으로 딕셔너리 요소(dictionary element)를 획득하도록 함으로써, 즉 순수한 암세포로부터 획득한 VOC를 이용하도록 함으로써, 환자간의 다양성(heterogenicity)을 극복하고자 한다. As described above, the present invention seeks to overcome the heterogeneity between patients by measuring VOCs of lung cancer cell lines and obtaining dictionary elements based on them, that is, by using VOCs obtained from pure cancer cells.
단계 S3의 세포 SERS 신호의 딥 러닝 단계에서는 학습 대상 세포 각각의 신호 패턴(즉, 특이적 SERS 피크)를 학습하도록 한다. In the deep learning step of the cell SERS signal of step S3, a signal pattern (ie, a specific SERS peak) of each cell to be learned is learned.
참고로, 딥 러닝의 일례로 딕셔너리 러닝(Dictionary learning)는 인공지능을 이용한 분석 방법의 일종으로, 다음과 같이 수행된다. For reference, as an example of deep learning, dictionary learning is a kind of analysis method using artificial intelligence and is performed as follows.
먼저, 도 6의 (a)와 같이 특정 데이터베이스는 아톰(atom)이라고 불리는 기본 요소들의 선형 조합인 딕셔너리(dictionary)로 표현한다. 그리고 데이터베이스에 존재하지 않는 새로운 신호(input)가 입력되면, 도 6의 (b)와 같이 새로운 신호(input)를 사전 내에 있는 atom들의 선형 조합(x=α1·d1 + α2·d2 + …+ αn·dn)(이때, w는 선형 계수)으로 표현한 후, 도 6의 (c)와 같이 최대한 많은 계수들이 0이 되게 하는 스파스 계수를 찾아내고, 이를 기반으로 딕셔너리(D)에서 새로운 신호(input)와 가장 비슷한 특성값을 가지는 신호를 파악하는 학습 알고리즘이다. 이때, atom은 데이터베이스에 포함된 신호 각각으로부터 도출되는 m개 신호 특성을 열 벡터(m×1)로 표현한 것으로, 이하에서는 설명의 편이를 딕셔너리 요소로 지칭하기로 한다. 딕셔너리 러닝은 딥 러닝의 일례로서 딕셔너리 요소는 SERS 신호의 신호 패턴으로 일반화되어 설명될 수 있다. First, as shown in FIG. 6A, a specific database is represented as a dictionary, which is a linear combination of basic elements called atoms. When a new signal that does not exist in the database is input, as shown in FIG. 6 (b), the linear combination of atoms in the dictionary with the new signal (x = α 1 · d 1 + α 2 · d 2 +… + Α n · d n ), where w is a linear coefficient, and then finds a sparse coefficient such that as many coefficients as 0 become as shown in FIG. 6 (c) and based on the dictionary D ) Is a learning algorithm that identifies the signal with the most similar characteristic value to the new signal. In this case, atom represents m signal characteristics derived from each of the signals included in the database as a column vector (m × 1), and hereinafter, the description will be referred to as a dictionary element. Dictionary learning is an example of deep learning, where a dictionary element can be described generalized to the signal pattern of the SERS signal.
이에 본 발명에서는 폐암 세포들과 정상 세포 각각의 SERS 신호(X)를 딥 러닝(deep learning)하여 세포 SERS 신호 각각에 대응되는 신호 패턴을 획득 및 저장하도록 한다. Therefore, the present invention deep learning the SERS signal (X) of each of the lung cancer cells and normal cells to acquire and store a signal pattern corresponding to each of the cell SERS signals.
단계 S4의 환자 호기 액화 단계는 도 7에서와 같이 수행될 수 있다. Patient exhalation liquefaction step of step S4 may be performed as in FIG.
즉, 환자 호기를 테들러 백(tedler bag)에 포집한 후(S41), 테들러 백에 실리콘 오일을 주입하고(S42), 기 설정 시간(예를 들어, 30분)이 경과하면, 테들러 백에 주사기를 삽입하여 환자 호기에 포함된 VOC가 용출된 실리콘 오일(즉, 호기 용출액)을 테들러 백으로부터 뽑아내도록 한다(S43). That is, after collecting the patient exhalation in the tedler bag (S41), injecting silicone oil into the Tedler bag (S42), when the preset time (for example, 30 minutes) elapsed, Tedler A syringe is inserted into the bag to allow the VOC contained in the patient's exhalation to remove the eluted silicone oil (ie, the exhalation eluent) from the Tedlar bag (S43).
단계 S5의 호기 SERS 신호 측정 단계에서는, 단계 S43에서 뽑아낸 호기 용출액을 SERS 기판에 공급한 후, SERS 기판에 785 nm 레이저를 조사하여 호기의 SERS 신호를 측정한다. In the step of measuring the exhalation SERS signal of step S5, after supplying the exhalation eluate extracted in step S43 to the SERS substrate, the SERS substrate of the exhalation is measured by irradiating the SERS substrate with a 785 nm laser.
단계 S6의 호기 SERS 신호 딥 러닝 단계에서는, 호기 SERS 신호에 대한 딥딕셔너리 러닝을 수행하여 호기의 신호 패턴(즉, 환자유래 SERS 신호 패턴)을 획득하도록 한다. In step S6, deep breathing of the exhaled SERS signal, a deep dictionary run on the exhaled SERS signal is performed to acquire a signal pattern of exhalation (ie, a patient-derived SERS signal pattern).
마지막으로 단계 S7의 폐암 세포 존재 여부 확인 단계에서는, 호기 SERS 신호의 신호 패턴과 단계 S3을 통해 학습된 신호 패턴간의 상관 분석을 통해, 도 8에서와 같이 호기 SERS 신호와 가장 유사한 신호 특성값을 가지는 세포를 검출한다. 도 8에서 붉은색 박스로 표시된 부분이 실제 환자유래 SERS 신호 패턴과 딥 러닝을 통해 학습된 SERS 신호 패턴의 상관관계가 가장 높은 세포의 쌍을 보여준다. Finally, in the step of checking whether lung cancer cells exist in step S7, through correlation analysis between the signal pattern of the exhaled SERS signal and the signal pattern learned through step S3, as shown in FIG. Detect the cells. The red box in FIG. 8 shows a pair of cells having the highest correlation between the actual patient-derived SERS signal pattern and the SERS signal pattern learned through deep learning.
그리고 검출된 세포가 폐암 세포이면, 사용자 신체에 폐암 세포가 존재함을 확인 및 통보하고, 검출된 세포 종류가 정상 세포이면, 사용자 신체에 폐암 세포가 존재하지 않음을 확인 및 통보하도록 한다. 또 다르게는 검출된 세포가 폐암 세포인 경우에 한해 폐암을 진단 및 통보하도록 한다. And if the detected cells are lung cancer cells, it is confirmed and notified that the lung cancer cells exist in the user's body, and if the detected cell type is normal cells, and confirms and notified that there is no lung cancer cells in the user's body. Alternatively, lung cancer may be diagnosed and notified only if the detected cells are lung cancer cells.
더하여, 상기의 방법에서는 딥 러닝의 일례로 딕셔너리 러닝을 들어 설명하였지만, 본 발명에서는 CNN(convolutional neural network)에 기반한 분석 동작 또한 지원할 수도 있도록 한다. In addition, in the above method, dictionary learning has been described as an example of deep learning, but the present invention may also support analysis operations based on a convolutional neural network (CNN).
도 9 및 도 10은 본 발명의 다른 실시예에 따른 호기 기반 폐암 진단 방법을 개략적으로 설명하기 위한 도면이다. 이는 딕셔너리 러닝 대신에 CNN에 기반한 방법이며, SERS 기판, SERS 신호 측정 등에 관련된 단계는 딕셔너리 러닝 기반 방법과 동일한 방식으로 수행되도록 하고, 이에 대한 상세한 설명은 생략하기로 한다. 9 and 10 are diagrams for schematically explaining the exhalation-based lung cancer diagnostic method according to another embodiment of the present invention. This is a CNN-based method instead of dictionary learning, and the steps related to SERS substrate, SERS signal measurement, etc. are performed in the same manner as the dictionary learning based method, and a detailed description thereof will be omitted.
CNN은 하나의 입력층, 다수의 은닉층, 하나의 출력층으로 구성되며, 각 층에 포함된 뉴런이 가중치를 통해 연결되며, 이에 세포 SERS 신호를 입력 조건으로, 세포 종류를 출력 조건으로 가지는 다수의 학습 데이터를 생성 및 학습시켜 각 층에 포함된 뉴런 가중치를 조정하도록 한다. CNN is composed of one input layer, multiple hidden layers, and one output layer, and neurons included in each layer are connected by weight, so that multiple learning with cell SERS signal as input condition and cell type as output condition Generate and train the data to adjust the neuron weights included in each layer.
본 발명에서는 이러한 CNN을 이용하여 세포 SERS 신호 또는 호기 SERS 신호의 신호 패턴을 파악하도록 하며, 특히 세포 종류를 통보하는 출력층 대신 은닉층의 i개(i는 2 이상의 자연수, 예를 들어, 10개)의 출력 정보를 입력 신호에 대응되는 신호 패턴으로써 획득 및 활용하도록 한다. In the present invention, the CNN is used to determine the signal pattern of the cellular SERS signal or the exhaled SERS signal, and in particular, i (i is 2 or more natural numbers, for example, 10) of the hidden layer instead of the output layer for notifying the cell type. The output information is acquired and utilized as a signal pattern corresponding to the input signal.
먼저, 도 9에 도시된 바와 같이 폐암 세포들과 정상 세포 각각의 SERS 신호(즉, 세포 SERS 신호)를 획득한다. First, as shown in FIG. 9, SERS signals (ie, cellular SERS signals) of lung cancer cells and normal cells are acquired.
그러면 세포 SERS 신호를 입력 조건으로, 세포 종류를 출력 조건으로 가지는 다수의 학습 데이터를 생성한 후 이들을 통해 CNN 모델을 반복 학습시켜, CNN 모델이 차후 입력 신호를 폐암 세포와 정상 세포로 이종 분류(binary classification)할 수 있도록 한다. CNN 모델이 학습 완료되면, 세포 SERS 신호 각각에 대한 i개(예를 들어, 10개)의 은닉층 출력 정보를 획득하고, 이를 세포 SERS 신호 각각의 신호 패턴으로 저장하도록 한다(S3‘). Then, after generating a large number of training data having cell SERS signal as input condition and cell type as output condition, CNN model is repeatedly trained through them, and CNN model subsequently classifies input signals into lung cancer cells and normal cells. classify. When the CNN model is trained, i (e.g., 10) hidden layer output information for each of the cellular SERS signals are obtained and stored as a signal pattern of each of the cellular SERS signals (S3 ').
단계 S3‘가 완료된 상태에서, 호기 SERS 신호가 획득되면(S4,S5), 호기 SERS 신호를 학습 완료된 CNN 모델에 입력한 후, 세포 SERS 신호와 마찬가지로 i개(예를 들어, 10개)의 은닉층 출력 정보를 호기 SERS 신호의 신호 패턴으로 획득 및 저장하도록 한다(S6‘). When the step S3 'is completed, when the aerobic SERS signal is obtained (S4, S5), the aerobic SERS signal is inputted into the trained CNN model, and then i (e.g., 10) hidden layers as in the cellular SERS signal. The output information is acquired and stored as a signal pattern of the exhalation SERS signal (S6 ').
그러고 나서, 세포 SERS 신호 각각의 신호패턴과 호기 SERS 신호의 신호 패턴간 거리를 계산하여, 가장 짧은 거리값을 가지는 세포 SERS 신호를 기반으로 호기 신호와 가장 유사도가 높은 세포를 도출 및 통보하도록 한다. Then, the distance between the signal pattern of each of the cellular SERS signal and the signal pattern of the aerobic SERS signal is calculated to derive and notify the cells having the highest similarity to the exhalation signal based on the cellular SERS signal having the shortest distance value.
가 어느 세포와 더욱 유사한지 거리 기반으로 판단하도록 한다. 이때, 거리 계산법은 마할라노비스(Mahalanobis) 거리, 유클리디안(Euclidean) 거리, 코사인(cosine) 거리 등 중 어느 하나일 수 있으며, 거리에 반비례하여 유사도는 증가되는 특징을 가진다. Determine distance based on which cells are more similar. In this case, the distance calculation method may be any one of Mahalanobis distance, Euclidean distance, cosine distance, etc., and the similarity is increased in inverse proportion to the distance.
이러한 방법을 통해 개별 호기 신호에서 나타나는 유사도를 계산할 수 있으며, 가장 바람직하게는 도 10에서와 같이 한 사람 당 50~250개의 호기 신호를 얻어 분석에 이용할 수 있도록 한다. 그리고 각 호기 신호별 유사도를 오름차순으로 정리하면, 환자 환자별로 폐암 세포와의 유사도를 손쉽게 도출할 수 있게 된다. Through this method it is possible to calculate the similarity appearing in the individual exhalation signal, and most preferably, 50 to 250 exhalation signals per person can be obtained and used in the analysis as shown in FIG. In addition, if the similarity for each exhalation signal is arranged in ascending order, the similarity with lung cancer cells can be easily derived for each patient.
도 11은 본 발명의 일 실시예에 따른 호기 기반 폐암 진단 시스템을 설명하기 위한 도면이다. 11 is a view for explaining a breath-based lung cancer diagnostic system according to an embodiment of the present invention.
도 11을 참고하면, 본 발명의 시스템은 SERS 기판(10), 상기 SERS 기판 각각에 공급된 세포 VOC(Volatile Organic Compound) 용출액들과 호기 VOC 용출액에 대한 SERS 신호를 측정하는 라만 분광기(20), 상기 라만 분광기에 의해 세포 SERS 신호들이 측정되면, 상기 세포 SERS 신호들의 패턴을 획득 및 저장하고, 호기 SERS 신호가 측정되면, 상기 호기 SERS 신호의 패턴을 획득 및 출력하는 딥 러닝부(30), 상기 호기 SERS 신호의 패턴과 상기 딥 러닝으로 학습된 패턴과의 상관 분석을 수행함으로써, 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 폐암 진단부(40) 등을 포함할 수 있다. Referring to FIG. 11, the system of the present invention includes a SERS substrate 10, a Raman spectrometer 20 for measuring SERS signals for cellular VOC (Volatile Organic Compound) eluates and aerobic VOC eluates supplied to each of the SERS substrates, When the cellular SERS signals are measured by the Raman spectrometer, the deep learning unit 30 acquires and stores the pattern of the cellular SERS signals, and when the exhaled SERS signal is measured, the deep learning unit 30 that acquires and outputs the pattern of the exhaled SERS signals, By performing correlation analysis between the pattern of the exhaled SERS signal and the pattern learned by the deep learning, the cell SERS signal having the highest similarity is detected, and the presence or absence of lung cancer cells is determined according to the type of the detected cell SERS signal. Lung cancer diagnosis unit 40 and the like.
이때, 세포 VOC 용출액은 세포를 배양한 배양액에 실리콘 오일을 주입하고 기 설정 시간 동안 교반한 후 상기 세포에 포함된 VOC가 용출된 실리콘 오일을 분리시킴으로써 획득 가능하다. 또한 호기 VOC 용출액은 환자 호기를 테들러 백(tedler bag)에 포집하고, 상기 테들러 백에 실리콘 오일을 주입하고 기 설정 시간 동안 기다린 후, 상기 환자 호기에 포함된 VOC가 실리콘 오일을 분리시킴으로써 획득 가능하다. At this time, the cell VOC eluate can be obtained by injecting silicone oil into the culture medium in which the cells are cultured and stirring for a predetermined time, and then separating the silicone oil eluted with VOC contained in the cells. Exhaled VOC eluate is also obtained by capturing the patient exhalation in a tedler bag, injecting silicone oil into the tedler bag and waiting for a preset time, after which the VOC contained in the patient exhalation separates the silicone oil. It is possible.
더하여, 이때의 세포도 앞서 설명한 바와 같이 폐암 세포와 정상 세포일 수 있으며, 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 중 적어도 하나일 수 있으며, 정상 세포는 Alveolar type II cell일 수 있다. In addition, the cells at this time may also be lung cancer cells and normal cells as described above, lung cancer cells are A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, It may be at least one of EBC1, H1650, N417, the normal cell may be an Alveolar type II cell.
즉, 본 발명의 시스템 또한 앞서 설명된 SERS 기술과 인공지능을 이용한 분류 방법인 딥 러닝(즉, 딕셔너리 러닝, CNN)을 이용하며, 이에 따라 비침습적이고 방사능 노출 걱정없는 폐암 진단 동작을 수행할 수 있도록 한다.That is, the system of the present invention also uses deep learning (ie, dictionary learning, CNN), which is a classification method using SERS technology and artificial intelligence described above, and thus can perform lung cancer diagnosis operation without worrying about invasive radiation exposure. Make sure
더하여, 상기에서는 폐암 세포와 정상 세포를 기반으로 폐암 여부만을 진단하는 경우에 한정하여 설명하였지만, 필요한 경우 기관지암, 결장 직장암, 전립선암, 유방암, 췌장암, 위암, 난소암, 방광암, 뇌암, 갑상선암, 식도암, 자궁암, 간암, 신장암, 담도암, 및 고환암 등 각각에 대응되는 암 세포를 기반으로 폐암 이외의 다양한 암도 진단할 수 있음은 물론 당연할 것이다. In addition, the above description is limited only to the diagnosis of lung cancer based on lung cancer cells and normal cells, but if necessary, bronchial cancer, colorectal cancer, prostate cancer, breast cancer, pancreatic cancer, gastric cancer, ovarian cancer, bladder cancer, brain cancer, thyroid cancer, Various cancers other than lung cancer may also be diagnosed based on cancer cells corresponding to esophageal cancer, uterine cancer, liver cancer, kidney cancer, biliary tract cancer, and testicular cancer.
이상의 설명은 본 발명의 기술 사상을 예시적베으로 설명한 것에 불과한 것으로서, 본 발명이 속하는 기술분야에서 통상의 지식을 가진 자라면 본 발명의 본질적인 특성에서 벗어나지 않는 범위에서 다양한 수정 및 변형이 가능할 것이다. 따라서, 본 발명에 개시된 실시예들은 본 발명의 기술 사상을 한정하기 위한 것이 아니라 설명하기 위한 것이고, 이러한 실시예에 의하여 본 발명의 기술 사상의 범위가 한정되는 것은 아니다. 본 발명의 보호 범위는 아래의 청구범위에 의하여 해석되어야 하며, 그와 동등한 범위 내에 있는 모든 기술 사상은 본 발명의 권리범위에 포함되는 것으로 해석되어야 할 것이다.The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

Claims (14)

  1. SERS(Surface Enhanced Raman Spectroscopy) 기판을 제작하는 단계;Fabricating a Surface Enhanced Raman Spectroscopy (SERS) substrate;
    다수의 세포 각각에 포함된 VOC(Volatile Organic Compound)를 용출하고, 세포 VOC 용출액 각각을 상기 SERS 기판에 공급한 후 세포 SERS 신호 각각을 측정하는 단계;Eluting a VOC (Volatile Organic Compound) contained in each of a plurality of cells, supplying each of the cell VOC eluates to the SERS substrate and measuring each cell SERS signal;
    상기 세포 SERS 신호 각각을 딥 러닝하여 상기 세포 각각의 신호 패턴을 학습하는 단계; Deep learning each cell SERS signal to learn a signal pattern of each cell;
    환자 호기를 포집한 후 실리콘 오일을 통해 액화시키고, 상기 액화된 환자 호기를 상기 SERS 기판에 공급하여 호기 SERS 신호를 측정한 후, 상기 딥 러닝 결과를 통해 분석하여 호기의 신호 패턴을 파악하는 단계; 및 Capturing a patient breath and liquefying with silicone oil, supplying the liquefied patient breath to the SERS substrate, measuring a breath SERS signal, and analyzing the deep learning result to determine a signal pattern of the breath; And
    상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴을 비교 분석하여, 상기 호기 SERS 신호와 가장 유사도가 높은 상기 세포 SERS 신호를 파악하고, 이를 기반으로 폐암 세포 존재 여부를 확인 및 통보하는 단계를 포함하는 호기 기반 폐암 진단 방법.Comparing the signal pattern of each of the cellular SERS signal and the signal pattern of the exhaled SERS signal to identify the cell SERS signal having the highest similarity to the exhaled SERS signal, and based on this to identify and notify the presence of lung cancer cells Exhalation-based lung cancer diagnostic method comprising the steps.
  2. 제1항에 있어서, 상기 다수의 세포는 The method of claim 1, wherein the plurality of cells
    폐암 세포와 정상 세포로 구성되며, 상기 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 중 적어도 하나일 수 있으며, 상기 정상 세포는 Alveolar type II cell인 것을 특징으로 하는 호기 기반 폐암 진단 방법.Lung cancer cells and normal cells, the lung cancer cells are at least one of A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 And, the normal cell is a breath-based lung cancer diagnostic method, characterized in that Alveolar type II cells.
  3. 제1항에 있어서, 상기 세포 SERS 신호 각각을 측정하는 단계는The method of claim 1, wherein measuring each of the cellular SERS signals comprises
    학습 대상 세포를 배양한 배양액에 실리콘 오일을 주입하고 기 설정 시간 동안 교반하여, 상기 학습 대상 세포에 포함된 VOC가 실리콘 오일에 용출되도록 하는 것을 특징으로 하는 호기 기반 폐암 진단 방법.Injecting silicone oil into the culture medium cultured the target cells and stirred for a predetermined time, the expiration-based lung cancer diagnostic method, characterized in that the VOC contained in the target cells to be eluted in the silicone oil.
  4. 제1항에 있어서, 상기 호기 SERS 신호를 측정하는 단계는The method of claim 1, wherein measuring the exhaled SERS signal comprises:
    환자 호기를 테들러 백(tedler bag)에 포집한 후, 상기 테들러 백에 실리콘 오일을 주입하고 기 설정 시간 동안 기다림으로써, 상기 환자 호기에 포함된 VOC가 실리콘 오일에 용출되도록 하는 것을 특징으로 하는 호기 기반 폐암 진단 방법.After collecting the patient exhalation in a tedler bag, by injecting silicone oil into the tether bag and waiting for a predetermined time, the VOC contained in the patient exhalation is eluted in the silicone oil Exhalation-based lung cancer diagnostic method.
  5. 제1항에 있어서, The method of claim 1,
    상기 세포 각각의 신호 패턴을 학습하는 단계는 딕셔너리 러닝 기반으로 상기 세포 SERS 신호 각각의 신호 패턴을 학습하고, Learning the signal pattern of each of the cells, learning the signal pattern of each of the cell SERS signal on the basis of dictionary learning,
    상기 호기의 신호 패턴을 파악하는 단계는 딕셔너리 러닝 기반으로 상기 호기 SERS 신호의 신호 패턴을 파악하는 것을 특징으로 하는 호기 기반 폐암 진단 방법.The step of identifying the signal pattern of the exhalation exhalation-based lung cancer diagnostic method, characterized in that for determining the signal pattern of the exhaled SERS signal based on the dictionary running.
  6. 제5항에 있어서, 상기 폐암 세포 존재 여부를 확인 및 통보하는 단계는 The method of claim 5, wherein the checking and notifying of the presence of lung cancer cells comprises
    상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴에 대한 상관 분석을 통해 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 것을 특징으로 하는 호기 기반 폐암 진단 방법.Correlation analysis between the signal pattern of each of the cellular SERS signals and the signal pattern of the exhaled SERS signal detects the cellular SERS signal having the highest similarity, and confirms the presence or absence of lung cancer cells according to the type of the detected cellular SERS signal. Exhalation-based lung cancer diagnostic method characterized in that.
  7. 제1항에 있어서, The method of claim 1,
    상기 세포 각각의 신호 패턴을 학습하는 단계는 하나의 입력층, 다수의 은닉층, 하나의 출력층으로 구성되는 CNN(Convolutional Neural Networks)를 통해 세포 SERS 신호와 세포 종류간의 상관 관계를 학습시킨 후, 학습 완료된 CNN를 통해 상기 세포 SERS 신호 각각의 신호 패턴을 획득하고, Learning the signal pattern of each cell is to learn the correlation between the cell SERS signal and the cell type through CNN (Convolutional Neural Networks) consisting of one input layer, a plurality of hidden layers, one output layer, Obtain a signal pattern of each of the cellular SERS signals via CNN,
    상기 호기의 신호 패턴을 파악하는 단계는 상기 학습 완료된 CNN를 통해 상기 호기 SERS 신호의 신호 패턴을 파악하되, Determining the signal pattern of the exhalation step is to determine the signal pattern of the exhalation SERS signal through the learned CNN,
    상기 세포 SERS 신호 각각과 상기 호기 SERS 신호의 신호 패턴은 상기 은닉층의 i개(i는 2 이상의 자연수) 출력 정보에 기반하여 결정되는 것을 특징으로 하는 호기 기반 폐암 진단 방법.The signal pattern of each of the cellular SERS signal and the exhaled SERS signal is determined based on i (i is a natural number of 2 or more) output information of the hidden layer.
  8. 제7항에 있어서, 상기 폐암 세포 존재 여부를 확인 및 통보하는 단계는 The method of claim 7, wherein the step of confirming and notifying the presence of lung cancer cells
    상기 세포 SERS 신호 각각의 신호 패턴과 상기 호기 SERS 신호의 신호 패턴간 거리값에 기반하여 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 것을 특징으로 하는 호기 기반 폐암 진단 방법.The cell SERS signal having the highest similarity is detected based on the distance between the signal pattern of each of the cell SERS signals and the signal pattern of the exhaled SERS signal, and the presence or absence of lung cancer cells is determined according to the type of the detected cell SERS signal. Exhalation-based lung cancer diagnostic method characterized in that.
  9. SERS(Surface Enhanced Raman Spectroscopy) 기판;Surface Enhanced Raman Spectroscopy (SERS) substrates;
    다수의 세포 각각의 VOC(Volatile Organic Compound) 용출액이 상기 SERS 기판에 공급되는 경우에는 세포 SERS 신호들을 측정하고, 호기 VOC 용출액이 상기 SERS 기판에 공급되는 경우에는 호기 SERS 신호를 측정하는 라만 분광기;A Raman spectrometer for measuring cellular SERS signals when a VOC (Volatile Organic Compound) eluate of each of a plurality of cells is supplied to the SERS substrate, and measuring an expiratory SERS signal when an exhaled VOC eluate is supplied to the SERS substrate;
    상기 라만 분광기에 의해 세포 SERS 신호들이 측정되면, 상기 세포 SERS 신호들에 기반한 딥 러닝을 수행하여 상기 세포 SERS 신호들의 신호 패턴을 획득 및 저장하고, 호기 SERS 신호가 측정되면, 상기 립 러닝 결과를 통해 상기 호기 SERS 신호의 신호 패턴을 획득 및 출력하는 딥 러닝부; 및When the cellular SERS signals are measured by the Raman spectrometer, deep learning based on the cellular SERS signals is performed to acquire and store signal patterns of the cellular SERS signals, and when the exhaled SERS signal is measured, the lip learning results A deep learning unit configured to acquire and output a signal pattern of the exhaled SERS signal; And
    상기 호기 SERS 신호의 신호 패턴과 상기 상기 세포 SERS 신호들의 신호 패턴을 비교 분석하여, 가장 높은 유사도를 가지는 세포 SERS 신호를 검출하고, 상기 검출된 세포 SERS 신호의 종류에 따라 폐암 세포 존재 여부를 확인하는 폐암 진단부를 포함하는 호기 기반 폐암 진단 시스템.Comparing the signal pattern of the exhaled SERS signal and the signal pattern of the cellular SERS signal to detect the cell SERS signal having the highest similarity, and to determine the presence of lung cancer cells according to the type of the detected cell SERS signal Exhalation-based lung cancer diagnostic system comprising a lung cancer diagnostic unit.
  10. 제9항에 있어서, 상기 세포 VOC 용출액은 The method of claim 9, wherein the cell VOC eluate is
    세포를 배양한 배양액에 실리콘 오일을 주입하고 기 설정 시간 동안 교반한 후 상기 세포에 포함된 VOC가 용출된 실리콘 오일을 분리시킴으로써 획득 가능한 것을 특징으로 하는 호기 기반 폐암 진단 시스템.Inhalation-based lung cancer diagnostic system, characterized in that by injecting a silicone oil in the culture medium cultured cells and stirred for a predetermined time, by separating the silicone oil eluted VOC contained in the cells.
  11. 제10항에 있어서, 상기 다수의 세포는 The method of claim 10, wherein the plurality of cells
    폐암 세포와 정상 세포일 수 있으며, 상기 폐암 세포는 A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 중 적어도 하나일 수 있으며, 상기 정상 세포는 Alveolar type II cell인 것을 특징으로 하는 호기 기반 폐암 진단 시스템.Lung cancer cells and normal cells, the lung cancer cells may be at least one of A549, H2087, H446, H460, H520, H358, H441, H2170, H157, H1299, H23, Calu-3, H522, EBC1, H1650, N417 And, the normal cell is a breath-based lung cancer diagnostic system, characterized in that Alveolar type II cells.
  12. 제9항에 있어서, 상기 호기 VOC 용출액은 10. The method of claim 9, wherein the exhaled VOC eluate is
    환자 호기를 테들러 백(tedler bag)에 포집하고, 상기 테들러 백에 실리콘 오일을 주입하고 기 설정 시간 동안 기다린 후, 상기 환자 호기에 포함된 VOC가 실리콘 오일을 분리시킴으로써 획득 가능한 것을 특징으로 하는 호기 기반 폐암 진단 시스템.After collecting the patient exhalation in a tedler bag, injecting silicone oil into the tedler bag and waiting for a predetermined time, the VOC contained in the patient exhalation can be obtained by separating the silicone oil Exhalation-based lung cancer diagnostic system.
  13. 제9항에 있어서, 상기 딥 러닝부는 The method of claim 9, wherein the deep learning unit
    딕셔너리 러닝 기반으로 딥 러닝하여 입력 신호의 신호 패턴을 획득 및 저장하는 것을 특징으로 하는 호기 기반 폐암 진단 시스템.Exhalation-based lung cancer diagnostic system, characterized in that the deep learning based on dictionary learning to acquire and store the signal pattern of the input signal.
  14. 제9항에 있어서, 상기 딥 러닝부는 The method of claim 9, wherein the deep learning unit
    하나의 입력층, 다수의 은닉층, 하나의 출력층으로 구성되는 CNN(Convolutional Neural Networks) 기반으로 딥 러닝하여 입력 신호 각각의 신호 패턴을 획득 및 저장하되, 상기 신호 패턴은 상기 은닉층의 i개(i는 2 이상의 자연수) 출력 정보에 기반하여 결정되는 것을 특징으로 하는 호기 기반 폐암 진단 시스템.Deep learning based on CNN (Convolutional Neural Networks) consisting of one input layer, a plurality of hidden layers, and one output layer to acquire and store a signal pattern of each input signal, i. 2 or more natural number) exhalation-based lung cancer diagnostic system, characterized in that determined based on the output information.
PCT/KR2019/006018 2018-05-18 2019-05-20 Exhalation-based lung cancer diagnosis method and system WO2019221581A1 (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
CN201980032939.7A CN112136036B (en) 2018-05-18 2019-05-20 Lung cancer diagnosis method and system based on expiration
US17/054,270 US20210247382A1 (en) 2018-05-18 2019-05-20 Exhalation-based lung cancer diagnosis method and system
EP19803582.6A EP3795983A4 (en) 2018-05-18 2019-05-20 Exhalation-based lung cancer diagnosis method and system

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
KR20180057191 2018-05-18
KR10-2018-0057191 2018-05-18
KR10-2019-0058785 2019-05-20
KR1020190058785A KR102225543B1 (en) 2018-05-18 2019-05-20 Method and system for lung cancer diagnosis based on exhaled breath

Publications (1)

Publication Number Publication Date
WO2019221581A1 true WO2019221581A1 (en) 2019-11-21

Family

ID=68540508

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2019/006018 WO2019221581A1 (en) 2018-05-18 2019-05-20 Exhalation-based lung cancer diagnosis method and system

Country Status (1)

Country Link
WO (1) WO2019221581A1 (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1225976B1 (en) * 1999-11-05 2010-06-16 Real-Time Analyzers, Inc. Material for surface-enhanced raman spectroscopy, and ser sensors and method for preparing same
WO2016184987A1 (en) * 2015-05-19 2016-11-24 Yaya Diagnostics Gmbh Means and methods for the enrichment of nucleic acids
WO2017021023A1 (en) * 2015-08-06 2017-02-09 Yaya Diagnostics Gmbh Means and methods for the detection of targets
EP3264070A1 (en) * 2016-06-30 2018-01-03 Sightline Innovation Inc. Waveguide-based system and method for biomarker detection

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1225976B1 (en) * 1999-11-05 2010-06-16 Real-Time Analyzers, Inc. Material for surface-enhanced raman spectroscopy, and ser sensors and method for preparing same
WO2016184987A1 (en) * 2015-05-19 2016-11-24 Yaya Diagnostics Gmbh Means and methods for the enrichment of nucleic acids
WO2017021023A1 (en) * 2015-08-06 2017-02-09 Yaya Diagnostics Gmbh Means and methods for the detection of targets
EP3264070A1 (en) * 2016-06-30 2018-01-03 Sightline Innovation Inc. Waveguide-based system and method for biomarker detection

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
CHEN ET AL.: "Breath Analysis Based on Surface-Enhanced Raman Scattering Sensors Distinguishes Early and Advanced Gastric Cancer Patients from Healthy Persons", ACS NANO, vol. 10, no. 9, 13 July 2016 (2016-07-13), pages 8169 - 8179, XP055654111 *

Similar Documents

Publication Publication Date Title
WO2020180003A1 (en) Artificial intelligence-based method and system for provision of information on cancer diagnosis by using exosome-based liquid biopsy
Fdez et al. Cross-subject EEG-based emotion recognition through neural networks with stratified normalization
Liu et al. Gut microbiota as an objective measurement for auxiliary diagnosis of insomnia disorder
US20130034910A1 (en) Diagnosing, prognosing and monitoring multiple sclerosis
Winholtz et al. Vocal tremor analysis with the vocal demodulator
Gitter et al. Trans/paracellular, surface/crypt, and epithelial/subepithelial resistances of mammalian colonic epithelia
US11604133B2 (en) Use of multi-frequency impedance cytometry in conjunction with machine learning for classification of biological particles
CN102342858A (en) Chinese medicine sound diagnosis acquisition and analysis system
WO2017056493A1 (en) Cancer development risk assessment device, program, and method for testing cancer development risk
CN111297403B (en) Rapid and accurate screening and early warning system for pulmonary fibrosis lesion of pneumoconiosis group
WO2019221581A1 (en) Exhalation-based lung cancer diagnosis method and system
Xi et al. Exhaled aerosol pattern discloses lung structural abnormality: a sensitivity study using computational modeling and fractal analysis
Chen et al. Latent and abnormal functional connectivity circuits in autism spectrum disorder
CN114155879A (en) Abnormal sound detection method for compensating abnormal perception and stability by using time-frequency fusion
KR102225543B1 (en) Method and system for lung cancer diagnosis based on exhaled breath
WO2021172852A2 (en) Device and method for calculating stroke volume using ai
Hildebrand et al. Nondispersive infrared spectrometry: a new method for the detection of Helicobacter pylori infection with the 13C-urea breath test
Iltis et al. Simultaneous dual-plane, real-time magnetic resonance imaging of oral cavity movements in advanced trombone players
CN115472293A (en) Lung adenocarcinoma multiomic diagnosis model based on serum metabolic fingerprint and construction method thereof
Klumpp et al. The phonetic footprint of covid-19?
Li et al. Flexible multivariable sensor based on mxene and laser-induced graphene for detections of volatile organic compounds in exhaled breath
WO2024053903A1 (en) Electronic nose sensor for diagnosing lung cancer, and electronic nose system for diagnosing lung cancer using same
Gunn et al. Using the psychological stress evaluator in conditions of extreme stress
WO2024096165A1 (en) Artificial intelligence-based mental illness diagnosis system and method using exosome sers signals
Li et al. Application of visual sensing technology in lung cancer screening

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19803582

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2019803582

Country of ref document: EP

Effective date: 20201218